PoWER-BERT: Accelerating BERT Inference via Progressive Word-vector Elimination

1 code implementation • ICML 2020 • Saurabh Goyal, Anamitra R. Choudhury, Saurabh M. Raje, Venkatesan T. Chakaravarthy, Yogish Sabharwal, Ashish Verma

We demonstrate that our method attains up to 6. 8x reduction in inference time with <1% loss in accuracy when applied over ALBERT, a highly compressed version of BERT.

General Classification Natural Language Understanding

Efficient Inferencing of Compressed Deep Neural Networks

no code implementations • 1 Nov 2017 • Dharma Teja Vooturi, Saurabh Goyal, Anamitra R. Choudhury, Yogish Sabharwal, Ashish Verma

Large number of weights in deep neural networks makes the models difficult to be deployed in low memory environments such as, mobile phones, IOT edge devices as well as "inferencing as a service" environments on cloud.

Quantization

Fenchel Duals for Drifting Adversaries

no code implementations • 23 Sep 2013 • Suman K. Bera, Anamitra R. Choudhury, Syamantak Das, Sambuddha Roy, Jayram S. Thatchachar

Existing literature shows (nearly) optimal drifting regret bounds only for the $\ell_2$ and the $\ell_1$-norms.